Kernel methods for machine learning - WMM9MO14

Goal(s)

Introduction to statistical learning theory and kernel-based methods.
Applications in bioinformatics, computer vision, text mining, audio processing, etc.

Content(s)

I. Introduction

I.1. Statistical learning: issues and goals
I.2. Risk convexification and capacity control
I.3. Convex optimization for statistical learning
I.4 Real applications

II. Kernel-based methods

II.1. Similarity measures and reproducing kernels
II.2. Reproducing kernel Hilbert spaces
II.4. Main families of reproducing kernels
II.3. Regularization as spectral function

III. Supervised statistical learning

III.1. Kernel Ridge Regression
III.2. Kernel Logistic Regression
III.3. Support Vector Machine
III.4. Capacity control and risk bounds

IV. Unsupervised statistical learning

II.1. Kernel Principal Component Analysis
II.2. Kernel Canonical Correlation Analysis
II.3. Spectral clustering
II.4. Large margin clustering
III.4. Capacity control and risk bounds

Probability, statistics, linear algebra.